Awareness

[Bill_Williams1]

CSGfolk,

In the 14 November issue of Science there is an article, "The Architecture
of Cognitive Control in the Human Prefrontal Cortex." p. 1181-85. The authors
Etienne Koechlin et al are a Paris based research group. I won't attempt an evaluation of the paper, except to say that it looks like a somewhat confusedly formulated control theory model. At least it has control loops of a sort. The paper also talks about stimulus-and-response links.

There is no mention of Powers.

Bill Williams
  UMKC

[Richard_Marken2]

[From Rick Marken (2003.12.04.2200)]

Bruce Gregory (2003.12.04.1730)--

The following is to be taken LITERALLY.

OK. I'll play. Though I can't believe that you can't answer these
questions yourself by now.

What exactly is "wanting" in PCT?

"Wanting" is an informal description of the role of the reference
signal in a control loop. A reference signal specifies the wanted state
of a perceptual variable. "I want a new car" could be translated into
"My reference for the state of the perceptual variable 'type of car I
drive' is 'new' ".

Can I want something without controlling for it?

Yes, if you consider controlling for an imagined version of the
perception to be not actually controlling (since you are not acting on
the world to get the perceptual variable into the wanted state).

Could that lead to conflict?

Not if you are controlling for the imagined version of the perception.
Otherwise, yes, it could.

Can the U.S. perceive
itself as liberating Iraq while Iraqis perceive the U.S. as occupying
Iraq?

Certainly.

The Iraqis want the occupiers out, but U.S. sees no occupiers.
Now an outside observer can say they are both controlling the same
perception the presence of the U.S. in Iraq

If they are both controlling for the presence of the U.S in Iraq then
the outside observer would be correct. However, the fact that one
person (or group of people) sees the U.S. as liberators and another
sees them as occupiers says nothing about whether either party is
controlling for the presence of the US in Iraq. Though I think that,
in fact, this has a lot to do with it (see the analysis below).

, but from the Iraqi and U.S. point of view it that obvious? Or even
true?

It's neither obvious nor necessarily true. The way to test it is to
observe how each party deals with the perception of the U.S. presence
in Iraq. Based on what I observe (via news reports) it looks to me
like some Iraqi's and most U.S. troops are controlling the same
perceptual variable -- call it "the location of U.S. forces" --
relative to two different references. Some Iraqi's seem to want this
variable in the state "out of Iraq" and some U.S. troops (or, at least,
their commanding officers) want this variable in the state "in Iraq".
The variable "the location of U.S. forces" cannot be in two different
states -- "in Iraq" and "out of Iraq" -- at the same time. So both
parties are acting against each other -- with increasingly deadly force
-- to get the perception of "the location of U.S. forces" into two
mutually exclusive states.

The reason why the two parties want "the location of U.S. forces" in
two different states may result from the fact that they are controlling
different higher level perceptions. The U.S. commanders may be
controlling for keeping the perceived level of a perceptual variable
that might be called "freedom in Iraq" in the state "high". Many
Iraqi's may be controlling for keeping another perceptual variable, one
that might be called "foreign occupation" in the state "none". The U.S.
commanders would then be controlling for Iraqi freedom by wanting U.S
forces in Iraq. The Iraqis would be controlling for no foreign
occupation by wanting U.S forces out of Iraq.

But it is not necessary that the two parties to a conflict be
controlling different higher level perceptual variables ("freedom in
Iraq" and "foreign occupation" in the example) for the conflict over
the lower level perceptual variable -- "the location of U.S. forces"--
to exist. All that is needed is that the two parties set different
references for the same lower level variable as the means of
controlling the higher level perceptions. For example, it's possible
that both parties are controlling the same higher level perceptual
variable -- "freedom in Iraq" -- relative to the same reference --
"high". What produces the conflict is that one party controls for this
higher level perception by setting a reference for "the location of
U.S. forces" to "in Iraq" and the other party controls for the same
perception by setting a reference for "the location of U.S. forces" to
"out of Iraq".

Best regards

Rick

···

---
Richard S. Marken
marken@mindreadings.com
Home 310 474-0313
Cell 310 729-1400

[Bruce_Gregory4]

[From Bruce gregory (2003.12.05.0658)]

Rick Marken (2003.12.04.2200)

Bruce Gregory (2003.12.04.1730)--

What exactly is "wanting" in PCT?

"Wanting" is an informal description of the role of the reference
signal in a control loop.

PCT definition noted.

Can I want something without controlling for it?

Yes, if you consider controlling for an imagined version of the
perception to be not actually controlling (since you are not acting on
the world to get the perceptual variable into the wanted state).

So if I want to retire but am doing nothing about retiring, I am
controlling in imagination for being retired. Sounds strained, but PCT
explanation noted.

So one approach to the conflict is not to attempt to change the goals
of higher level perceptions. but rather the lower level perceptions
both parties use to achieve those higher level goals.

Bruce Gregory

"Everything that needs to be said has already been said. But since no
one was listening, everything must be said again."

                                                                                Andre Gide

[Bill_Powers]

[From Bill Powers (2003.12.05.0746 MST)

Bruce Gregory (2003.12.04.1730)

What exactly is "wanting" in PCT?

I think I agree with Rick in wondering why, after all this time, you won't
try to answer this question yourself.

We use the term "want" in two ways. That ought to be enough of a hint.

Best,

Bill P.

[Richard_Marken2]

[From Rick Marken (2003.12.05.0850)]

Bruce Gregory (2003.12.05.0658)--

So if I want to retire but am doing nothing about retiring, I am
controlling in imagination for being retired. Sounds strained, but PCT
explanation noted.

Your love/hate conflict over PCT is showing. If what you say sounds strained perhaps it's
because you created a strained sounding definition of "controlling in imagination" to
attribute to PCT. A better definition of "controlling in imagination", I think, is "wanting
to retire and perceiving, in imagination, what it would be like to be retired, whether one is
currently acting to do something to produce that perception in actuality (by putting money in
a 401K, for example) or not. Of course, the best definition of "controlling in imagination"
and other aspects of PCT is provided by the PCT model itself. Therefore, one of the best
definitions of "controlling in imagination" is provided by my spreadsheet model, which let's
you "flip the imagination switch" for a control system by entering an "*" above that system
so you can see what happens when one control system in a hierarchy controls in imagination.

So one approach to the conflict is not to attempt to change the goals
of higher level perceptions. but rather the lower level perceptions
both parties use to achieve those higher level goals.

Yes. If either party to the conflict were able to change the lower level controlled variable
so that controlling it both satisfied their higher level goal and didn't interfere with the
other party's control of what had been the same lower level controlled variable, then you
have solved the conflict. In the Iraq example, this would require that one party stop
controlling for the perception of the location of U.S. troop as the means of controlling
their higher level perception. If, for example, the U.S. started controlling for Iraqi
freedom by other means than having U.S. troops present in Iraq -- for example, by having an
international force present -- then the Iraqi's could control for not being occupied by
keeping the perception of U.S. troop presence at zero with no resistance from the U.S., since
the U.S. would no longer be controlling that perception.

Best regards

Rick

···

---
Richard S. Marken, Ph.D.
Senior Behavioral Scientist
The RAND Corporation
PO Box 2138
1700 Main Street
Santa Monica, CA 90407-2138
Tel: 310-393-0411 x7971
Fax: 310-451-7018
E-mail: rmarken@rand.org

[Bruce_Gregory4]

[From Bruce Gregory (2003.12.05.1320)]

[From Bill Powers (2003.12.05.0746 MST)

Bruce Gregory (2003.12.04.1730)

What exactly is "wanting" in PCT?

I think I agree with Rick in wondering why, after all this time, you
won't
try to answer this question yourself.

Because when I do try myself I tend to draw dismissive if not outright
hostile responses. In addition, I never would have guessed the PCT
story about controlling in the imagination mode for things I want but
am not actively controlling for achieving. The best I can say for this
explanation is that it is unlikely to be challenged by data. My
simpler, apparently non-PCT explanation, is that gain associated with
achieving such wants is sufficiently low that I never get around to
pursuing them.

Bruce Gregory

"Everything that needs to be said has already been said. But since no
one was listening, everything must be said again."

                                                                                Andre Gide

[Bruce_Gregory4]

[From Bruce Gregory 92003.12.05.1335)]

Rick Marken (2003.12.05.0850)]

Bruce Gregory (2003.12.05.0658)--

So if I want to retire but am doing nothing about retiring, I am
controlling in imagination for being retired. Sounds strained, but PCT
explanation noted.

Your love/hate conflict over PCT is showing.

I'm going to have to do a better job of hiding that love/hate business.

So one approach to the conflict is not to attempt to change the goals

of higher level perceptions. but rather the lower level perceptions
both parties use to achieve those higher level goals.

Yes. If either party to the conflict were able to change the lower
level controlled variable
so that controlling it both satisfied their higher level goal and
didn't interfere with the
other party's control of what had been the same lower level controlled
variable, then you
have solved the conflict. In the Iraq example, this would require that
one party stop
controlling for the perception of the location of U.S. troop as the
means of controlling
their higher level perception. If, for example, the U.S. started
controlling for Iraqi
freedom by other means than having U.S. troops present in Iraq -- for
example, by having an
international force present -- then the Iraqi's could control for not
being occupied by
keeping the perception of U.S. troop presence at zero with no
resistance from the U.S., since
the U.S. would no longer be controlling that perception.

This suggests a "rule" of conflict resolution. Seek the lowest level at
which a change of goals can be brought about while still maintaining
higher-level (hence more difficult to change) goals.

Bruce Gregory

p.s. it seems to me that lowering the gain in a control loop has
similar effects to controlling in imagination.

"Everything that needs to be said has already been said. But since no
one was listening, everything must be said again."

                                                                                Andre Gide

[bnhpct]

[From Bruce Nevin (2003.12.05 14:25 EST)]

Bruce Gregory (2003.12.02.1900)--

(Awareness is after all purposeful.)

How is awareness purposeful?

It seems that I can control what I am aware of. But in that case, some hypothetical control system is what is purposeful, and what it is controlling has the effect that I call awareness. I won't speculate here about the mechanism, the point is that it is not awareness that is purposeful.

Suppose I focus my awareness on some perception. I soon find my attention wandering elsewhere. If I persist in paying attention to that chosen perception, I find it difficult to do anything else competently. "Paying attention" to that perception seems to mean "ignoring" other perceptions. (Perhaps this indicates adjustment of gain higher on the chosen perception and lower on the others.) On the other hand, if I stop and do nothing else but attend to that chosen perception, the "nothing else" turns out to be densely populated, and it is not long before my attention wanders off with one control process or another. All this is a commonplace of meditation practice, of course -- but is awareness itself purposeful?

Maybe I should ask what you mean by purposeful here.

         /Bruce Nevin

···

At 07:02 PM 12/2/2003 -0500, Bruce Gregory wrote:

[bnhpct]

[From Bruce Nevin (2003.12.05 15:40 EST)]

Bruce Gregory (2003.12.03.0621)--

Martin Taylor 2003.12.13.0046 EST

It has always seemed to me that conscious awareness happens under two
circumstances, which might resolve to just one after due
consideration. The two are: (1) some control system encounters
persistent error and is unable to bring the controlled perception
near its reference value, and (2) there is a possibility of shifting
from controlling one perception to controlling another, where there
would be conflict (in the sense recently accepted) if one were to try
to control both at once.

It may be a bit of a stretch to include the pleasurable observation
of a sunset under case 2, but I believe it does fit there. I could be
persuaded that there's a case 3 that better fits the sunset
observation, or that case 2 needs some expansion. But for now, that's
my intuition about awareness.

Mine, too. Thanks.

Purposeful direction of attention to a particular perception comes under (1) soon enough after one starts meditating, but not at the outset. Perhaps (3) is during acquisition of control.

It is unclear to me what is controlled when we listen to one sound to the exclusion of other concurrent sounds. The proposal that some control system adjusts the relative gain of other systems seems to lead to a great complication of the hierarchy.

Explanations for wanting something but not actively controlling it:
1. Controlling in imagination mode (Rick Marken 2003.12.05.0850)
2. Lowering the gain so you "never get around to it" (Bruce Gregory 2003.12.05.1320, 92003.12.05.1335)
3. Controlling unsuccessfully (Bruce Nevin 2003.12.05 12:25 EST)
4. Controlling something not yet available for control (Bruce Nevin 2003.12.05 12:25 EST) -- cat at mouse hole, bully at door -- or by means that are not yet available and must be waited for. This is a special case of (3).
5. Controlling at some point in a sequence, at the end of which is the controlled result can give the appearance of not controlling the end point.

There is no a priori reason one of these must be exclusively true and the others incorrect.

On (2) there is much unclarity, at least in my mind, about control systems adjusting the gain of other control systems.

Number (5) seems to me to explain why one does not go as fast as possible at the beginning of a trip. (This was based on the assumption that gain increases with error -- the "arrive at destination" error is at its greatest then.) A long-term goal, such as retirement, is controlled by intermediate steps, not all of which can be controlled at a given moment (4) and not all of which are controlled with the same gain as one's control of the endpoint.

Number (5) also might help explain deceleration during a trip. Or, rather, provide context for the explanation that has been given. A substep to avoid an accident (under a control system with high gain) does not in itself immediately conflict with the end goal of arriving at the destination. And as Bill said the immediate steps that you were engaged in (continue driving east on the Pike; pass that car, avoid traffic at that lane drop) are controlled at much lower gain.

         /Bruce Nevin

···

At 06:22 AM 12/3/2003 -0500, Bruce Gregory wrote:

[Marc_Abrams1]

from [Marc Abrams (2003.12.05.1552)]

[From Bruce Nevin (2003.12.05 14:25 EST)]

How is awareness purposeful?

You seem to be speaking of awareness as if it were actually a 'thing' That
is an object, something that you can actually seperate from other things and
look at independently. I think it might be more useful to look at it as an
emergent state of several interacting processes, with control being a
contributing but not a neccesary component of one of those interacting
systems.

It seems that I can control what I am aware of.

Sometimes. Can you control the speed and direction of cars other than your
own that you are aware of on the road?

Can you stop the pain after hitting your 'funny' bone in the elbow?

But in that case, some hypothetical control system is what is purposeful,

and what it is

controlling has the effect that I call awareness. I won't speculate here
about the mechanism, the point is that it is not awareness that is

purposeful.

I think there might be a tendency to confound two things here. One can have
a 'purpose' (i.e. a goal) and try to attain it. This usually involves many
control processes over time. The 'other' purpose is relegated to a single
control process that is _only_ concerned with minimizing error for its very
specific control purpose, _regardless_ of the 'purposes' or 'goals' the
larger system may have. Awareness needs to exist on both levels but they are
of a different nature. At a higher level, it's cognitive. At a lower level
it might be the gradient of a chemical.

Suppose I focus my awareness on some perception.

You focus your 'attention' not your awareness. But this seems to be a simple
mistatement based on your comments that follow

I soon find my attention wandering elsewhere. If I persist in paying

attention to that chosen

perception, I find it difficult to do anything else competently. "Paying
attention" to that perception seems to mean "ignoring" other perceptions.

yes, until an error occurs 'pulling' you in a different direction.

(Perhaps this indicates adjustment of gain higher on the chosen perception
and lower on the others.)

That would be my guess.

On the other hand, if I stop and do nothing else
but attend to that chosen perception, the "nothing else" turns out to be
densely populated, and it is not long before my attention wanders off with
one control process or another. All this is a commonplace of meditation
practice, of course -- but is awareness itself purposeful?

I don't think so.

Marc

[Bruce_Gregory4]

[From Bruce Gregory (2003.12.05.1628)]

[From Bruce Nevin (2003.12.05 15:40 EST)]

On (2) there is much unclarity, at least in my mind, about control
systems
adjusting the gain of other control systems.

I think this possibility should be excluded on the basis that it does
not play a role in orthodox HPCT.

Bruce Gregory

"Everything that needs to be said has already been said. But since no
one was listening, everything must be said again."

                                                                                Andre Gide

[Bill_Powers]

[From Bill Powers (2003.12.05.1452 MST)]

Bruce Gregory (2003.12.05.1628)--

[From Bruce Nevin (2003.12.05 15:40 EST)]

On (2) there is much unclarity, at least in my mind, about control
systems
adjusting the gain of other control systems.

I think this possibility should be excluded on the basis that it does
not play a role in orthodox HPCT.

If a person is still trying to grasp the model in B:CP, I would agree with
this advice. However, once the basic concept of control via reference
signals is understood, and the mechanisms of hierarchical control are
mastered, there's no reason to avoid proposals about controlling through
parameter adjustments in lower systems. I haven't gone into them because so
far I haven't done any experiments that need parameter adjustment in the
model. Others are welcome to think up such experiments and see where they lead.

One hint: an effect of having too high a loop gain is to make control
systems start to oscillate spontaneously. I know there are situations where
I must relax on purpose to avoid shakiness -- and that I can do this
relaxing and cure the shakes. So I can perceive a consequence of too high a
loop gain, and do something about it, which suggests that parameter control
is a distinct likelihood.

Best,

Bill P.

[Richard_Marken2]

[From Rick Marken (2003.12.05.1400)]

Bruce Gregory 92003.12.05.1335)--

This suggests a "rule" of conflict resolution. Seek the lowest level at
which a change of goals can be brought about while still maintaining
higher-level (hence more difficult to change) goals.

I think this rule is OK. But I think you could be clearer about what seeking the lowest level
means. The lowest level relevant to a conflict is the level at which the conflict is
occurring. That is, it's the level at which the incompatible goals are set for the same
perception. Changing goals at this level basically means adding a degree of freedom to the
one currently available for control of perceptions at the next higher level. But it's not
always physically possible to do this. For example, suppose you are in conflict about wanting
to go to the store and wanting to go to the movies. The conflict here results from having two
goals for one variable (df) -- your location. This is the lowest level of the conflict. You
could solve this conflict by changing the one location goal into two, which would basically
mean splitting yourself in two. Of course, this won't work so you can't solve this conflict
at the lowest level -- the level of the conflict itself.

I think it's probably rare that conflicts can be solved by adding degrees of freedom at the
level of the conflict . I think what usually has to happen is this: Once you've found the
lowest level of the conflict -- the level at which the conflict exists in the sense described
above -- you have to bring your awareness up a level , to the level that is setting the
incompatible goals for the single degree of freedom (your location, in the example) and see
if there is a way to set _those_ goals (the goals of going shopping and to the movies, in the
example) so that there is no conflict. I think this is the way conflicts are usually resolved
because there usually _is_ a way to set the higher level goals so that the conflict
disappears. It's done by sequencing the goals: you want to shop _before_ you want to go to
the movies. Since you can be in two different places at two different times, the conflict no
longer exists.

So I think a possibly better way to describe your rule of conflict resolution is 1) identify
the conflict (the lowest level perception that is wanted in two states) 2) see if it's
possible to add a perception to control at that level. If so, the conflict is solved. If not
3) try to become aware of the higher level goals responsible for wanting the single variable
(degree of freedom) in two different states and 4) change how those higher level goals are
set.

I think that's kind of what the MOL is about.

Best regards

Rick

···

--
Richard S. Marken, Ph.D.
Senior Behavioral Scientist
The RAND Corporation
PO Box 2138
1700 Main Street
Santa Monica, CA 90407-2138
Tel: 310-393-0411 x7971
Fax: 310-451-7018
E-mail: rmarken@rand.org

[Richard_Marken2]

[From Rick Marken (2003.12.05.1550)]

Bill Powers (2003.12.05.1452 MST)--

Bruce Gregory (2003.12.05.1628)--

>I think this possibility [gain adjustment -- RM] should be excluded on the
basis that it does
>not play a role in orthodox HPCT.

If a person is still trying to grasp the model in B:CP, I would agree with
this advice. However, once the basic concept of control via reference
signals is understood, and the mechanisms of hierarchical control are
mastered, there's no reason to avoid proposals about controlling through
parameter adjustments in lower systems. I haven't gone into them because so
far I haven't done any experiments that need parameter adjustment in the
model. Others are welcome to think up such experiments and see where they lead.

I have Rx fulfillment error data that I account for using gain adjustment in a
two-level version of my prescribing error model. I need more data to test the gain
adjustment approach to fitting the model to data but right now it looks _very_
promising.

Best

Rick

···

--
Richard S. Marken, Ph.D.
Senior Behavioral Scientist
The RAND Corporation
PO Box 2138
1700 Main Street
Santa Monica, CA 90407-2138
Tel: 310-393-0411 x7971
Fax: 310-451-7018
E-mail: rmarken@rand.org